The Song Fight Bulletin Board

She Sells Sea Shells by the Sea Shore.

And recording this sentence sounds like 'sh-s-s-sh-s-s.'

Anyone have any good tips for making recordings that don't sound like that? And how to appropriately process things afterwards to get rid of the problems that remain?

I've tried using Logic's De-Esser plugin, but that just maketh me thound like I'm lithping, even if I only put it on the reverb thend. I also found some compressor settings that purport to do a better job, but it cautheth the thame problem.

Some recording forums recommend just recording vocals with the mic off-axis but all that does is muffle everything.

Rewriting my lyrics to avoid any 'S' sounds i jut not ac-eptible.

Back when I was using Cool Edit Pro/Adobe Audition, I relied heavily on their handy built-in spectral analysis tool. It changed the waveform into a spectral view, and you could just highlight the bright-yellow (indicating high-volume) section in the higher frequency area for the length of time when it's actually there, and reduce that by about 6db. It would leave the rest of the audio untouched. Probably the lazy way out, but I was usually satisfied with the results*. Depending on your DAW, there may very well be a plugin that can do this for you.

Since switching to Cubase, I have yet to find a better way to de-ess my vocal performances. I use a pop filter positioned several inches away from the condenser, and most of the time things tend to sound decent. If the sibilant sounds do become a disruption, I just try to re-record the audio, turning my head away from the microphone or backing up a step during the problem words.

If all else fails, I try spot-reducing the volume (cutting 5-10db of audio on just the sibilant noise, which is usually like .2 seconds, and then having it come right back up immediately). This isn't very elegant and I don't like it.

*Caveat: I had much lower standards back then; it's possible that I would hate it if I had access to the same tool today.

That select-and-attenuate method is just a fiddly manual version of doing exactly what Logic's de-esser plugin does - it does an EQ cut during particular characteristic spectral peaks. The sidechained single-band compressor approach is similar (except it cuts everything, not just the problem frequencies).

Really the problem is that there's too much crap in the signal in the first place, and I'd like to figure out a way of recording that doesn't introduce the problem to begin with.

You are probably already trying to use vocal techniques to cut down on the sibilance in your singing, right? For example:

I've tried thtuff like that but it jutht thoundth like a lithp. Got any other ideath?

The particular track I'm having trouble with is a rerecord of Strangers, which has a lot of problematic words in it (house, spouse, mouse).

The lady in that video doesn't sound like she's lisping. Maybe you're not doing it right and need to keep working on it? I mean, if a de-esser doesn't work, and that is a purpose built tool for doing exactly what you want, how do you expect some other processing technique to solve it better...?

Have you tried different mics? A cheap MXL condenser in my experience makes these issues worse. You might want to try a dynamic mic instead.

I'm not asking for a processing technique, I'm asking for recording techniques in general that will cut it down (mic placement, mouth shaping, etc.).

And by "all that stuff" I was referring to the general corpus of sibilance-reduction videos I found on YouTube. That particular video you embedded doesn't address the particular issue that I'm having in my recordings. I mean, yeah, she doesn't sound like she's lisping but she still has plenty of sibilance on the S sounds that aren't off an 'E' sound, and none of the problematic things in my song have an 'E' sound around the S either.

And yes I fully realize I'm doing something wrong! I'm trying to figure out what that is and fix it.

ken wrote:Have you tried different mics? A cheap MXL condenser in my experience makes these issues worse. You might want to try a dynamic mic instead.

Thought about it, but never got around to trying it, but I'll give it a shot. I prefer recording with a condenser but I have some dynamics on hand, and also some non-MXL condensers as well but they're all small-diaphragm. I haven't had this problem on other songs with this exact same mic but on the other hand I've never had a song where every other word rhymes with "house" either.

Okay, I tried a quick take with my ElectroVoice dynamic this morning and it sounds... well, better, although still not amazing. I'll try tooling around with that tonight.

It sounds to me that you're heading in the right direction and trying to get as much out of your recording as possible. I personally like dynamic mics, but they can sound muffled in the mid/upper range. So everyone EQ's for clarity by rolling off the bottom and boosting for brightness, then all of a sudden, sibilance out the yin-yang!

But that's not the answer. Because you either record a track with sibilance and you mix to take it out, or you record a track with minimal sibilance and by the time you mix it, you add sibilance back in. I personally like a condenser mic on lead vocals and a dynamic on backup vocals. It gives a nice separation in the frequency field and when mixed, sent through my chain and panned, very pleasing to my ears. Just a personal preference, of course.

The bottom line, you will almost always need a de-esser. I don't care how hard you try to record without sibilance. You will eventually sound held back and not natural. I'd rather hear esses than someone that sounds like they aren't going for broke as if there is a baby sleeping in the next room.

With that said, I've tried a number of de-essers. I ended up sticking with one that was free that gets used almost exclusively. It's called spitfish from the fish fillet suite. In fact I also use the floorfish in the suite quite often as well. http://www.digitalfishphones.com/main.p ... &subItem=5

The key is, if you are making the esses sound like th, you are trying to get rid of too much.
Here's some food for thought. Use the spitfish (or your de-esser) pre-set before you start tweaking it. You'll think it's too subtle, but hold back! Then render the song and listen. You'll be amazed at how well the sibilance flows and fits the finished song.

Another thing I do. Let's say I have the de-esser set and it's perfect for 99% of the vocals. But there is that one word that jumps out at me. I don't want to stop or there's no time to re-record. I put that word on it's own track and treat it independently. But I've had to go so heavy, the whole word sounds fucked. I'll then put the word back on the main vocal track, clip out the ess and ONLY treat the ess independently.

Also as a side note. I actually use the lisp sound from the de-esser on purpose. Spitfish has a pre-set called "cute girl from Sweden". I love using it on djDen's vox. It really does add a cute charm.

Hm, looks like they have AudioUnit versions. Awesome! I'll try it out. Thanks for the link.

And yeah I'm trying to just remove only a little bit but it seems like a big problem is my reverb is amplifying it. One of the things I tried doing was breaking the reverb out onto its own send and only de-essed the send, but that still had too much of a lithpy thound to it.

It almost sounds like you're trying to get rid of too much ess. Remember that getting rid of too much ess is just as bad as having too much. It sounds more natural with some sibilance. You might want to pop a reference track on next to your rendered song and give it an A-B back and forth test to hear how the sibilance is compared to yours. You might be over focusing on it and can't get your brain passed it. Like a faucet dripping when you're trying to sleep, lol.

Sure, I don't want to get rid of the S sounds, obviously, but I want to get rid of the "I hear nothing but S S S S S S S S" effect, wherein the harsh treble of the sibilance is reverberating like crazy and totally dominates the vocals. It's like the old SNL sketches with the Chicago folks who are talkin' 'bout Da Bearssss and Da Bullssss.

Ha, yeah, I know that sound. It grates.

Aw, dang, those plugins haven't been updated since 2005, and don't work on current Intel Macs. Phooey.

don't use so many s words in your lyrics.

PROBLEM SOLVED

NEXT

Somewhat shitty solution.

If your reverb seems like it's compounding the problem, you may want to pick a different one, or at least put a lowpass in front of it (only on the send, of course). I've found that that makes a huge difference with "cheap-sounding" reverbs, and so maybe it will make a difference in this case, too.

Yeah I've tried a whole bunch of different reverb treatments. The lowpass helped to cut down on the sibilance, but then made it sound muddy instead of bright. I know, it's a balancing act...

So far JB's suggestion seems to be the only workable one and this song just doesn't work with different words. :P

Okay, I used the following to get it to sound pretty okay (although now it's not quite as bright as I'd like, but that's the tradeoff):

• Rerecorded off-axis with a dynamic mic and kept my tongue a bit looser during the S sounds
• Gentle de-esser on the main output
• Low-pass filter on the reverb send
• Decreased reverb level

Before and after?

I have stopped the practice of keeping a billion copies of every mix and whatever, because it isn't productive to look back on how a track has changed over time. But you can preorder the album to hear the after as soon as it comes out. ;)

EDIT: Oh, I guess I can go back in my backup system to get a 'before' version. I don't like the direct comparison because it makes it even more obvious how relatively-muddy it is after, but here you go.